Build, Baby, Build. In Fact, Overbuild.

[Bob Wallace]

So often I’ve read people claim that wind and solar can’t be more than 10% of grid supply, that wind and solar can’t grow larger than their capacity factor, or that wind and solar must be 100% backed up by fossil fuel or storage.

I decided to see how far one could push wind and solar penetration at a reasonable cost and with no fossil fuel backup.  I’m using California 2017 data because it’s the only publicly available hourly data for load and wind/solar generation I’ve found.  Percent penetration = Total electricity supplied / Total demand.

The bottom line?  The route to a least expensive grid for California may be through overbuilding wind and solar generation rather than lesser amounts of wind and solar generation along with short term (one to three day) battery storage.

Since this is a ‘what might be in 20 years’ I’m assuming that both wind and solar will cost no more than $0.02/kWh (2018 dollars) to generate.  Both are already closing on that price with unsubsidized wind now below $0.03/kWh for several power purchase agreements (PPAs) and solar approaching $0.03/kWh in best cases. 

By 2035 a sizable portion of California wind and solar farms should have paid their capital and financing costs bringing their generation costs (operating costs) to less than $0.01/kWh.  This average of ~$0.02 new generation and <$0.01 paid off generation makes an assumed price of $0.02/kWh a very reasonable assumption.  (All costs in 2018 dollars.)

For this study I used CAISO 2017 hourly demand (load) along with actual wind and solar generation for each specific hours. 

The total demand for 2017 was 231 terawatt hours (TWh).  Actual solar generation during the year was 29 TWh.  Actual wind generation was 14 TWh, about half the amount generated from solar.

For each hour of the year I multiplied solar and wind production to find the highest penetration levels possible without driving the cost too high. 

While some have argued that wind and solar penetration will be limited by their capacity factor this is clearly not the case.  By simply increasing the amount of both wind and solar generated in 2017 five times from the year’s production penetration rises to more than 70%.  With 5x solar and 5x wind 40% of each separate hour would have seen its demand fully met by electricity direct from wind and solar farms.  No storage.

At 5x solar and 15x wind 73% of all hours would be fully supplied by only wind and solar.  Penetration would be 89%.




If electricity from wind and solar farms that costs $0.02/kWh then what would it cost to achieve high penetration levels with only wind and solar straight from turbines and panels with no storage involved?   89% penetration for $0.03/kWh at 5x solar and 15x wind. 



Overbuilding creates a lot of unused (surplus/curtailed) generation.  The more unused potential generation the higher the cost of electricity.  Solar panels and wind turbines are spending time sitting idle and not earning money.  At 5x solar and 15x wind California could generate 125 TWh more electricity than the grid demands.




The solution to high wind and solar penetration and lower electricity costs is to find other uses for the unused generation.  An obvious use is charging EVs.

In 2014 Californians drove over 350 billion miles.  At 0.28 kWh/mile that works out to 268,762 MWh per day or 98 TWh per year.  Let’s assume that charging times can be controlled by the grid, either directly or via price signals, and all EVs are recharged each day as much as possible with potential generation the grid does not need.

The following table shows the number of days per year on which all EVs could not be charged with available overgeneration.  Once the number of days drops below 20 it should be possible to avoid fully charging some EVs on that specific day and charge them only enough to provide the following day’s needs.



From this point on I’m restricting the amount of solar and wind to 15x for each.  There may be a better (cheaper) combination but at this point I’m not seeking the best, but to illustrate the role of  overbuilding.

At 15x solar and 15x wind penetration is almost 90%.  On fewer than 10 days a year are not all EVs fully charged.  And the cost of electricity drops from $0.07/kWh to $0.05/kWh due to sales for EV charging. 



More creative EV charging could allow for less overbuilding and lower the price.  Selling curtailed electricity for desalination, pumping water over the mountains to SoCal, or other dispatchable uses could further lower the cost.  But even at $0.05/kWh there is no lower cost low carbon option.
---

View this as an early draft.  Look for problems, please.  I'm not very good at self-editing.

Many, many thanks to Etienne for checking my spreadsheets (and finding some mistakes).  If anyone else would like to take a close look I'll be glad to share the link.

Obviously using one state and one year does not create a 'one size fits all' set of numbers.  But I think it gives us a strong suggestion that pure wind and solar may be usable for a high percentage of our electricity leaving a minor roll for other renewables and storage.

As for California and most 'west of the Rockies' states I suspect the penetration level for wind and solar will move close to 100% as we add offshore wind. 

[oren]

Great post. Note another obvious use for the extra electricity is desalination.
I do think that it's better to include some storage in the calculation, including hydro dams that can fill some of the "holes" in the coverage. The no storage assumption is severe.

[DrTskoul]

Who would build with the expectation of earning no money?

[johnm33]

Who would build with the expectation of earning no money?

There've been much worse make-work schemes.

[Sigmetnow]

Great post. Note another obvious use for the extra electricity is desalination.
I do think that it's better to include some storage in the calculation, including hydro dams that can fill some of the "holes" in the coverage. The no storage assumption is severe.

Thanks for crunching the numbers, Bob!
I agree that storage will become a big player over the next several years.  But I’m thinking of the rising popularity of batteries, both at a utility scale and residential grid-tied, in addition to smart EV charging.  Which only makes these numbers seem that much more feasible.

Edit:
Not just “popular.”  Mandated!

https://www.utilitydive.com/news/california-puc-finalizes-new-500-mw-btm-battery-storage-mandate/441901/

[DrTskoul]

Who would build with the expectation of earning no money?

There've been much worse make-work schemes.

All with massive government action and spending.... Which government is capable and willing to do such without a catastrophy right at the gates or a palpable existential fear...

[Bob Wallace]

Great post. Note another obvious use for the extra electricity is desalination.
I do think that it's better to include some storage in the calculation, including hydro dams that can fill some of the "holes" in the coverage. The no storage assumption is severe.

My goal was to see how far wind and solar could be 'pushed'.  If California did install 15x (or some other large amount) of wind and solar, reaching close to 90% penetration existing storage (CA has PuHS) and dispatchable hydro along with existing geothermal could fill the last ~10%.

Installing new storage in order to build less wind and solar doesn't seem to make economic sense.  We have no long term storage that costs less than $0.10/kwh.  It wouldn't make sense to spend ten cents to save five cents.

[Bob Wallace]

Who would build with the expectation of earning no money?

I'm not sure what you are asking.  Are you asking where the profits for the wind and solar farms are?

If so, I used expected PPA costs based on current PPAs and trends.  Power purchase agreements are contracts between owners of wind, solar, and other plants.  The contract sets a price the plant owner will receive for power delivered. 

The owner is going to include return on their investment building the facility, operating costs, and profits.  They won't sign a contract and build a plant for less.

I expect, based on current PPAs, the rate at which costs are falling, and the opinions of many in the industry that we are looking at roughly $0.02/kWh PPAs for both wind and solar within the next ten years. 

Those are likely to be locked down contracts which if signed at two cents will stay at two cents for the 20 to 25 year life of the contract.

[Bob Wallace]

But I’m thinking of the rising popularity of batteries, both at a utility scale and residential grid-tied, in addition to smart EV charging.  Which only makes these numbers seem that much more feasible.

My model uses one hour blocks.  The average draw and production during an hour.  We will need batteries or some other storage device to do the moment to moment supply/demand matching grids require.

Behind the meter storage might change the numbers somewhat but I'm not seeing a big impact coming from consumer storage.  What I've modeled, if implemented (and I've made no significant error), would eliminate the high cost some consumers in California encounter.  Five cents plus five more for distribution plus a couple more cents to administer the grid would bring CA electricity rates to about the twelve cent US average.  It probably would not make economic sense to install behind the meter storage, or solar, if electricity was that inexpensive.

[magnamentis]

don't forget about hydrogen production.

hydrogen can be stored and used for various appliances, last but not least fuel cells.

[Sigmetnow]

... What I've modeled, if implemented (and I've made no significant error), would eliminate the high cost some consumers in California encounter.  Five cents plus five more for distribution plus a couple more cents to administer the grid would bring CA electricity rates to about the twelve cent US average.  It probably would not make economic sense to install behind the meter storage, or solar, if electricity was that inexpensive.

But there are backup generators galore today, for hospitals, pumps, server farms....
Future needs will depend not just on grid energy cost, but criticality of the service, stability of the grid, frequency of severe weather that cuts grid power, etc.  The beauty of smart grid-tied storage backup (instead of diesel generators) is that even if the power never goes out, the batteries can be “used” by the grid for stabilization — and make money for the owner!

[Bob Wallace]

... What I've modeled, if implemented (and I've made no significant error), would eliminate the high cost some consumers in California encounter.  Five cents plus five more for distribution plus a couple more cents to administer the grid would bring CA electricity rates to about the twelve cent US average.  It probably would not make economic sense to install behind the meter storage, or solar, if electricity was that inexpensive.

But there are backup generators galore today, for hospitals, pumps, server farms....
Future needs will depend not just on grid energy cost, but criticality of the service, stability of the grid, frequency of severe weather that cuts grid power, etc.  The beauty of smart grid-tied storage backup (instead of diesel generators) is that even if the power never goes out, the batteries can be “used” by the grid for stabilization — and make money for the owner!

Backup generators mainly function to fill in when wires go down, not generation.  That need won't disappear and the batteries are probably not the solution.  We need to move backup generation to biofuels such as biodiesel or cleaner burning bio-methane.

[Sigmetnow]

Backup generators mainly function to fill in when wires go down, not generation.

No reason storage can’t do both, and lots of reasons why it should. (Grid balance, instant-on, multiple-use [not just sitting around rusting, waiting for a power outage that might not happen], no fuel storage requirements, minimal/no maintenance, no noise, provides a payback....)

[numerobis]

Behind the meter storage might change the numbers somewhat but I'm not seeing a big impact coming from consumer storage.

Behind-the-meter and consumer I agree. But I see a possible path for mass adoption of in-home storage, where the storage is connected to a central system to provide benefits to the grid most of the time (so it's not behind the meter, nor operated by the consumer, but is in the home/business). The virtual power plant model.

It would cost a bit more than batteries at substations and power plants. However, the consumer might be willing to pay the increment because they get backup power when the grid falls over.

That increment is a lot less than buying a system specifically for backup power. Some people and businesses already do that, so clearly there's demand even at a high price. Drop the price, demand will grow a lot.

The grid operator would save money on parts for the grid (smaller transformers and wires needed) and could continue billing even in a storm.

[Bob Wallace]

Backup generators mainly function to fill in when wires go down, not generation.

No reason storage can’t do both, and lots of reasons why it should. (Grid balance, instant-on, multiple-use [not just sitting around rusting, waiting for a power outage that might not happen], no fuel storage requirements, minimal/no maintenance, no noise, provides a payback....)

It costs a lot of money to store energy in batteries.  Batteries need to cycle often in order to earn money and pay for themselves. 

Energy in the form of liquid fuel can be stored for very little money.

We can't expect hospitals, for example, to pay vastly more for a very large battery bank to carry them through days of grid outage.  They can store enough for a short period and more fuel can be delivered if needed.

[Bob Wallace]

Behind the meter storage might change the numbers somewhat but I'm not seeing a big impact coming from consumer storage.

Behind-the-meter and consumer I agree. But I see a possible path for mass adoption of in-home storage, where the storage is connected to a central system to provide benefits to the grid most of the time (so it's not behind the meter, nor operated by the consumer, but is in the home/business). The virtual power plant model.

It would cost a bit more than batteries at substations and power plants. However, the consumer might be willing to pay the increment because they get backup power when the grid falls over.

That increment is a lot less than buying a system specifically for backup power. Some people and businesses already do that, so clearly there's demand even at a high price. Drop the price, demand will grow a lot.

The grid operator would save money on parts for the grid (smaller transformers and wires needed) and could continue billing even in a storm.

It's probably cheaper for the utility to disperse its storage around neighborhoods.  Purchase storage at wholesale rates.  Use the storage for grid smoothing as well as backup.

But we've wandered far off the topic of overbuilding. 

[Archimid]

Nice. As I read it I keep thinking of the duck curve and how it relates to over building.



Does your analysis differentiate between day hours and night hours? It seems to me that overbuilding of widely distributed wind might need to be much higher to cover afternoon peak hours and night hours.

I think this article (where the graph comes from) is very relevant to overbuild.

https://www.vox.com/energy-and-environment/2018/3/20/17128478/solar-duck-curve-nrel-researcher

FTA:

The consensus is emerging that we can probably do 80 percent [renewables] with some combination of spatial diversity and short-duration storage.

We can deal with diurnal shifts with short-duration storage, and not too much of it. When we did our Renewable Electricity Future study back in 2012, we got up to 80 percent renewables with only about 100 GW of additional storage. It’s not that much.

[TerryM]

I like the idea of the overbuild proposal. Built in redundancy when things do go south.


Storing energy as electricity, as opposed to water behind a dam, or even some of the more bizarre kinetic energy or pressurized air proposals seems to add large periodic expenses that the overbuilt solution skips over.
At some point every battery will need to be replaced. Every improperly maintained battery has to be seen as a fire hazard.


Terry

[Bob Wallace]

Does your analysis differentiate between day hours and night hours? It seems to me that overbuilding of widely distributed wind might need to be much higher to cover afternoon peak hours and night hours.

I can probably do an hour by hour analysis.  Let me think about the best way to set it up. The data is already spread across two separate sheets as 12 months exceeds the GooSheet max.

[Bob Wallace]

Doing an hour by hour may take some time.  I cranked solar down to 3x and wind to 5x in order for there to be enough unserved hours to see a change over the hours of the day.

Here's January.  First number is first hour of the day = 30 out of 744.  24 hours x 31 days.  Last number is 11 pm to 12 am = 31.  Fewest not covered days in the middle of the day.

30   30   31   31   31   31   31   31   17   10   8   7   6   7   7   16   31   31   31   31   31   31   31   31

For January wind needs to be cranked higher to cover more non-midday hours.  The '8' is the 10 am to 11 am block.

Looking at the way solar come in gradually and doesn't stay at peak for very long suggests to me that there's a lot of non-tracking solar in the mix.

Tracking yields a more square shouldered curve and a longer peak.



I'll do a summer month and see what that looks like.

[Bob Wallace]

June

Underserved hours are 12 am to 6am and 6 pm to midnight. More wind needed.

23   23   22   24   27   17   0   0   0   0   0   0   0   0   0   0   0   6   26   29   29   29   28   25

Does that answer your question or are you wondering about something else?

[Bob Wallace]

The duck lives.

(Removed this chart.  Didn't seem right.  Look below for new version.)

Obvious solution is to crank wind higher.

[Archimid]

Obvious solution is to crank wind higher.

Obviously.

Now I'm curious as to what it looks like at higher wind overbuild. Solar looks maxed out at 3x in June, the rest is EV juice. How many more multiples of wind to behead the duck once and for all?

[Bob Wallace]

Obvious solution is to crank wind higher.

Obviously.

Now I'm curious as to what it looks like at higher wind overbuild. Solar looks maxed out at 3x in June, the rest is EV juice. How many more multiples of wind to behead the duck once and for all?

Working on that.  Should have a flock of ducks put together by tonight.

[sidd]

I take it you are aware of Shaner(2018)

doi: 10.1039/c7ee03029k

sidd

[Bob Wallace]

I take it you are aware of Shaner(2018)

doi: 10.1039/c7ee03029k

sidd

Assuming minimal excess generation, lossless transmission, and no other generation sources, the analysis indicates that wind-heavy or solar-heavy U.S.-scale power generation portfolios could in principle provide ∼80% of recent total annual U.S. electricity demand. However, to reliably meet 100% of total annual electricity demand, seasonal cycles and unpredictable weather events require several weeks’ worth of energy storage and/or the installation of much more capacity of solar and wind power than is routinely necessary to meet peak demand. To obtain ∼80% reliability, solar-heavy wind/solar generation mixes require sufficient energy storage to overcome the daily solar cycle, whereas wind-heavy wind/solar generation mixes require continental-scale transmission to exploit the geographic diversity of wind.

First, when I see that one of the authors is Ken Calderia an alarm starts ringing in the background.  The guy has a hard-on for nuclear.  He's part of that POS critique of Jacobson's paper. (Stephen Davis also appears both places.)

Now, could all the world get 80% of its electricity from only wind and solar without storage?  I don't know.  I only have data for one year in California. I also have 3 years of wind and demand data for ERCOT.  It looks like ERCOT could get over 70% from wind alone with no storage at a good price via overbuilding.

Did the paper you link deal recognized that the answer to high wind and solar penetration is storage or overbuilding?

...to reliably meet 100% of total annual electricity demand, seasonal cycles and unpredictable weather events require several weeks’ worth of energy storage and/or the installation of much more capacity of solar and wind power than is routinely necessary to meet peak demand.

What they do not say that it is much cheaper to overbuild than to store.

Here's a recent tweet from Claderia -

If we discovered nuclear power today, we would be working like mad to make it as safe and cheap as possible.

As if we hadn't been working hard for over a half century to make nuclear affordable.  And we've failed.

Here are global unsubsidized average (mean) costs.  Look at 2017.



Solar and wind can be vastly overbuilt before they reach the cost of nuclear.

Nuclear, as soon as it exceeds minimum demand needs storage to shift its production around in time to match demand.  Or reactors have to start load following which also drives up the cost of electricity.  I've never seen a pro-nuclear person admit that without being forced to do so.

[Bob Wallace]

Some Duck Portraits.

I charted four levels of solar and wind for the month of January and June.  The "1xS 1xW" is actual 2017 generation.  The highest multiples I ran (5xS 15xW) knocked midday hours closer to but not all the way to zero.  This is the time of year when storage such as pump-up hydro might be brought into play.



At 1xS 1xW there were almost no June days fully supplied.  By 5xS 15xW almost all hours of the month are fully supplied.



I do expect this to greatly change when we start harvesting offshore wind.  We could easily have no problem supplying close to 100% of all hours with only wind and solar and with less than "15x wind". 

[numerobis]

Shaner et al has various scenarios. They all use storage. Their best scenario uses 50% overbuilding in solar and wind ("balanced" they call it).

The chief weakness is that it's a polemic paper: they set out to demonstrate that solar and wind alone isn't sufficient unless you drastically overbuild. They don't look at other generation methods like hydro or biomass. And they don't look at concepts like demand reduction (other than what's embedded in the demand data they use).

I read their paper as implying that we can pretty easily get to 100% renewables year-round.

[etienne]

I enjoyed very much working with Bob on his data files. One thing that he didn't say yet and that I found very interesting is that the Fall is the time of the year with the most problems to cover the load. It's quite logical, it is still hot, but days get shorter and weather is often not so good (thunderstorms...) At work, I have simulated PV self-consumption without storage on a facility storing fruits in Luxembourg and found similar results, there are some summer evenings with a low PV production and high energy need.

My personnal conculsions are in this context :
- Renewable can probably power us, but we need more than two sources (wind and solar in this example), and might have to use technologies like wood gazeification to produce electricity from renewable sources.
- Storage is probably the best solution for short time balance of the network, but long time storage is too expensive.Overbuilding is better. V2G might help a lot because most car batteries can be emptied in the evening and filled again after midnight. Same thing at noon if you eat lunch at work.
- Demand response is an issue. I like the comments RED made in the Renewable Energy Trend https://forum.arctic-sea-ice.net/index.php/topic,256.msg150443.html#msg150443 . With a efficient smart grid, curtailment could be really reduced, but this requires a lot of infrastructure change and an optimised strategy otherwise oscillations would be created (if you just tell to all the users in a town that electricity is now 20% cheaper, load will increase too much for you to handle it). Concepts are now available, like deep-freezers where the temperature can move between -20°C and -40°C so that cooling only happens when power is available, or water heater that can overheat, wash-machine with WIFI connection… but the smart grid concept is not ready excepted for basic demand-response. People using these technologies trie to optimise locally their energy use, for example because they are off-grid or have PV panels.
- I feel that technologies that will be developed to use power overproduction need to be extremely flexible, they also need to be cheap or have to work with a demand-responde scheme. I fell that overproduction is not stable enough for complex business models.

It’s not related but the problem is similar. I have seen a model using rain water to supply water in toilets for a big residential building. I was very surprised to see how stable the water availability was. I checked the data and found that if you do an average on a few years, water availability is very stable, but if you look the years separately, water availability is never twice the same on a specific month. Bob’s study has the advantage of using one year data, so results can be trusted.

[Bob Wallace]

Shaner et al has various scenarios. They all use storage. Their best scenario uses 50% overbuilding in solar and wind ("balanced" they call it).

The chief weakness is that it's a polemic paper: they set out to demonstrate that solar and wind alone isn't sufficient unless you drastically overbuild. They don't look at other generation methods like hydro or biomass. And they don't look at concepts like demand reduction (other than what's embedded in the demand data they use).

I read their paper as implying that we can pretty easily get to 100% renewables year-round.

If they surmise that we can reach 100% renewables then I assumed incorrectly after reading only the abstract.  I assumed, based largely on the pro-nuclear authors, that they made a "costs too much" argument.

Obviously we'd need large amounts of overbuilding or large amounts of storage.  I'm suggesting, based on the data I have, that it would be far less expensive, at least in California and Texas, to "massively" overbuild than to do very large amounts of storage.

Wind and solar have become inexpensive while storage remains comparatively expensive.

I was just reflecting this morning on my personal 'grid'.  For the last three days we've had bright sunny skies and my batteries have been full by 11am. 

That's after a winter in which I used my generator more than I like.  Many days I'd produce enough solar from cloudy skies to keep refrigerator, computers, and stereo running but I added nothing to the batteries to carry me through the evening and night.

I'd like to cut my generator use to as little as possible.  I simply cannot afford to purchase enough batteries to store large amounts of energy for days/weeks.  But I can afford to increase my solar array by 3x.

With 3x as much solar I might be able to generate 24  hours of electricity on moderately cloudy days and need to store or use the generator only on days when the clouds are dark. 

Overbuilding is the most promising solution for me.  And I suspect for everyone.

[Archimid]

Build, baby, build.

If we overbuild solar so that 30 minutes before sunset on a cloudy December 21, a base load is produced that exceeds ten times the maximum demand ever recorded, then during the day the grid becomes virtually a source of infinite energy. Sounds like prosperity to me.

How much overbuild we need for something like that? Is there even enough space on land?

[Bob Wallace]

Build, baby, build.

If we overbuild solar so that 30 minutes before sunset on a cloudy December 21, a base load is produced that exceeds ten times the maximum demand ever recorded, then during the day the grid becomes virtually a source of infinite energy. Sounds like prosperity to me.

How much overbuild we need for something like that? Is there even enough space on land?

I have not been able to get any data on the relationship between density of cloud cover and panel output. 

Some time back I calculated the area needed to power the US lower 48 with only PV solar.  The land requirement would be 0.4% of all land.

Let's assume that on the very worst day with the entire US covered by heavy cloud cover it might take 10x as many panels.  That would mean covering 4% of our land with panels.  But not all of the country is going to be covered with thick clouds at the same time.  We could install less than 10x and use transmission.

It could be done.  Put panels on most buildings, over parking lots, on brownfields and land ruined by coal mining, on burned out farmland.  But mixing in wind makes more sense.  In the end wind, especially as we start using offshore wind, is likely to play a larger role than solar.  A 50%/40% sort of distribution.

[numerobis]

Bob, I recommend you read the paper, not just the abstract. Basically they're running the same simulation you're running, but based on much more detailed data, taking a lot more into consideration than you do.

The ulterior motive you fear is indeed apparent in their conclusions (though they're a bit coy to go right out and say you need nuclear). Reading the paper does not force you to accept their conclusions. I don't -- I draw the opposite conclusion!

[Bob Wallace]

I'll read the paper but 50% overbuilding is not enough.  Based on the data I've been able to use.

Wind and solar are dropping to $0.02/kWh.  Nuclear and storage are over $0.10/kWh.

We can greatly overbuild 2 cent generation before we start getting close to the cost of storage or nuclear.

2 cents * 1.5 (50% overbuilding) = 3 cents.

2 cents * 5 (500% overbuilding) = 10 cents.

 

[Bob Wallace]

I don't think I'll be reading that paper.

I'm not going to spend 42.5 British pounds for it.

If you've got access how about checking to see if they suggest going higher than a 50% overbuild.  That's 50% over the point at which wind + solar provide 100% of demand for a single hour.  Not high enough.

I want to, I want to, I want to take you higher
I wanna take you higher
Baby, baby, baby, light my fire
I wanna take you higher

Boom laka-laka-laka
Boom laka-lak-goon-ka boom

[numerobis]

You can find the paper, e.g. by emailing the authors.

Here's a blog post that better explains the paper:
https://kencaldeira.wordpress.com/2018/03/01/geophysical-constraints-on-the-reliability-of-solar-and-wind-power-in-the-united-states/

Money quote:

All of these studies share common ground. They all indicate that lots more wind and solar power could be deployed today and this would reduce greenhouse gas emissions. Controversies about how to handle the end game should not overly influence our opening moves.

[etienne]

I finally found the abstract and somehow, they arrive to the same conclusion than I did, wind and solar won't make it alone, you don't need to analyse 36 years of data for that result. But there are so many other renewables, like hydro, wood, tides, waves, methanisation of biomass, biofuels... Than we have demand response, efficiency gains, and I believe that we wouldn't die if we would agree to reduce our comfort, couldn't we slow down trains, reduce AC in supermarket... if there is not enough PV power?

There is another aspect that is often forgotten, looking at the past when the world changes so much everyday might not provide specific information, but just a hint of what might happen. General conclusions are ok, but not specific ones. From Bob's work and from the abstract, we can tell that overbuilding is a part of the solution, that we need a lot of it, we can't tell how much we have to do it (15 to 20 times each) but it doesn't matter because we are far enough of the goal so adjustments are still possible before we reach it. We also can tell that demand-response will be needed, that long time storage is not a solution, that the issue is not closed and that creativity is needed everyday to go forward.

Thanks Numerobis for the link to the blog post. It's interesting, but nuclear is not really discussed. For me, nuclear and coal are not solutions also because they can't be curtailed, which is not compatible with a smart grid concept, more with a subsidized system.

[Bob Wallace]

My 15x each is 15x the amount California would to increase what they had in place in 2017. 

California is far from the "built" point.  During the first six months of 2017 (I'll add in more data later) the highest daily penetration was 75%.   If you define "built" as reaching peak demand for a single hour CA has a ways to go.

[crandles]

I'll read the paper but 50% overbuilding is not enough.  Based on the data I've been able to use.

Wind and solar are dropping to $0.02/kWh.  Nuclear and storage are over $0.10/kWh.

We can greatly overbuild 2 cent generation before we start getting close to the cost of storage or nuclear.

2 cents * 1.5 (50% overbuilding) = 3 cents.

2 cents * 5 (500% overbuilding) = 10 cents.

I agree overbuilding makes sense. Less sure about 500% overbuilding 10 cents plus cost of maintaining distribution makes electricity expensive.

But even well before 100% overbuild, who gets paid what? Does the price come down to practically zero virtually all the time so there is no incentive to continue to overbuild? If those with contract for supply get paid but those without don't, why would anyone without contract for supply invest in something when rarely get any income?

Is there a possible market mechanism? Maybe some feature where solar PV and turbine owners are paid to switch off generation until point where indifferent between earning for supply of electricity and earning for turning off generation? Can this work to produce a reasonably efficient outcome? Can it be implemented if some currently have contracts for supply?

[sidd]

I have mentioned two papers before on another thread:

99.9% of hours of load can be met by renewables with only 9-72 h of storage

90% of load can be supplied at an all up cost of 10c/Kwh. The numbers go to 99.9% and 20c/Kwh. We can already cut those costs by at least half.

doi:10.1016/j.jpowsour.2012.09.054

and

" ... when using future anticipated costs for wind and solar, carbon dioxide emissions from the US electricity sector can be reduced by up to 80% relative to 1990 levels, without an increase in the levelized cost of electricity. The reductions are possible with current technologies and without electrical storage.”

Need transmission build tho, for this one.

doi:10.1038/NCLIMATE2921

I made a more detailed comment at realclimate on these two papers:

http://www.realclimate.org/index.php/archives/2016/08/unforced-variations-aug-2016/comment-page-3/#comment-658616

sidd

[Bob Wallace]

I'll read the paper but 50% overbuilding is not enough.  Based on the data I've been able to use.

Wind and solar are dropping to $0.02/kWh.  Nuclear and storage are over $0.10/kWh.

We can greatly overbuild 2 cent generation before we start getting close to the cost of storage or nuclear.

2 cents * 1.5 (50% overbuilding) = 3 cents.

2 cents * 5 (500% overbuilding) = 10 cents.

I agree overbuilding makes sense. Less sure about 500% overbuilding 10 cents plus cost of maintaining distribution makes electricity expensive.

But even well before 100% overbuild, who gets paid what? Does the price come down to practically zero virtually all the time so there is no incentive to continue to overbuild? If those with contract for supply get paid but those without don't, why would anyone without contract for supply invest in something when rarely get any income?

Is there a possible market mechanism? Maybe some feature where solar PV and turbine owners are paid to switch off generation until point where indifferent between earning for supply of electricity and earning for turning off generation? Can this work to produce a reasonably efficient outcome? Can it be implemented if some currently have contracts for supply?

Let's go back to my study.  Had California 15x as much solar generation and 15x as much wind generation in 2017 as they actually had they could have 90% wind/solar penetration. 90% of all grid demand could have been covered by wind and solar direct from the farm.  And 100% of all light vehicles, assuming they were all EVs, could have been charged on all but six days of the year.

The cost to generate electricity, assuming a base cost for wind and solar of 2 cents, would be 4 cents.   A move from 2 to 4 cents indicates that about twice as much electricity is being generated as consumed.

(Finding other uses for the extra would bring the price down.)

As for the market mechanism, I don't know.  Right now we seem to be largely using PPAs where the amount to be delivered and price are contracted.  Perhaps later we'll see merchant wind and solar farms built.

The less of a wind or solar farm's output that is needed the more the farm will charge when it is needed.  That's what happens now with gas peakers.  They run less than 10% of the time but command high return when they are asked to perform.

[Bob Wallace]

Just some detail on how well California's 2017 installed wind and solar performed compared to demand.

The minimum hourly demand for the year was 18,296 MWh.  Wind and solar alone reached that level only 115 out of 8,760 hours.  The maximum demand of 49,900 was never reached.

So when I talk about 15xS 15xW that's off a base that is low penetration.

[Bob Wallace]

I'm going to drop this in here for comparison.  It's kind of long but I emboldened the critical sentence if you want only the most important bit.

It is imperative that we cease using fossil fuels and move to low carbon energy sources.  For electricity that means all nuclear energy, all renewable energy, or a combination of the two.  I have carried out a little study to see, at least at a basic level, what it might cost to create an all nuclear grid for CAISO (California Independent System Operator) members.  The combined four electricity grids in California.

I used 2017 California demand data that is available on the web.   I have similar demand data for ERCOT and will do a similar analysis on that data at a later date.  Obviously one year of data for one set of grids does not fit all.  But it should offer some insight into what would be required on a larger level.  For every grid there is a daily variation in demand and all grids operate with seasonal differences throughout the year.

This is ‘stage one’ of a 100% nuclear grid for California based on 2017 hourly loads and all current cars and light trucks becoming battery powered. 

Some basics about the model...

      1) “Reactors” are 1135 MW reactors such as the AP1000.  It is assumed the reactors would be able to load follow to some extent

      2)  Nuclear cost used is Lazard’s 2017 unsubsidized LCOE for new nuclear which ranges from $0.095 to $0.135/kWh.

      3) Cost of detailed load matching (integration costs) are not included.  Something such as batteries would be needed to match supply and demand on a finer grain level than reactors are capable of doing.

      4) The cost of backup for unscheduled reactor failure is not included.

Upon running the model it was found that it would take somewhere between 30 and 35 1135 MW reactors to produce enough electricity to supply hourly load without storage or other sources of electricity. 

That is 100% penetration and would mean that roughly 25% to 35% of the electricity produced would be unneeded (curtailed).  The cost of generating would be $0.18 to $0.20/kWh (plus backup and integration costs).



If all CA cars and light trucks were battery powered increasing the number of reactors to somewhere in the 40 to 45 count range all light vehicles could be charged each day, replacing the electricity used for that day’s driving. 



 The cost of electricity would be in the $0.16/kWh to $0.18/kWh range.  But that would be the cost of generation which is even less than the wholesale cost of electricity as it includes no cost for transmission or owner profit..  California’s retail rate for electricity is $0.15/kWh (Jan 2018).

It’s possible that adding some storage could decrease the number of reactors.  I’ll model that in later.  But with the Lazard median cost for PuHS being $0.175/kWh I can’t see storage helping.  It would take much less expensive storage. 

It’s also possible that selling some of the curtailed electricity for other uses such as desalinization would lower the cost but that would probably be more than offset by the costs not included.

Not included in the model is the cost of integrating large amount of nuclear onto the grid.  We can assume newly built reactors would be able to load follow to some extent but some amount of more flexible supply (probably batteries) would be needed for the second to second flexibility needed to maintain frequency and voltage control.

Plus there is the issue of reactors unexpectedly dropping off the grid.  I don’t have enough data to make a definitive  statement on the number of backup reactors which would be needed but some data on which to base a guess. 

Over a six month span beginning in September 2017, 17 of 98 or 17% of all US reactors dropped offline for reasons other than scheduled refueling and maintenance.  A greater than 30% failure rate on an annual basis.  The number of shutdowns may have been higher.  Sometimes it’s been months after the shutdown that I stumble over the news.

Some number of extra reactors would need to be running at reduced loads, ready to take up if one of the fleet dropped out.  And the reserve would need to be generous because sometimes one or more reactors can go offline for extended periods.  If a reactor is lost like, for example, Three Mile Island there needs to be a reserve reactor for permanent replacement.  We wouldn’t be able to wait five to ten years for a replacement to be constructed.

If we decided that we needed 40 reactors and 8 more for backup that would mean an increase of 20% in the cost of electricity.

In January 2018 the retail cost of electricity in California was $0.15/kWh.   Even at $0.16 for nuclear adding in $0.05 for distribution would drive the retail cost well over $0.20/kWh and have a significant impact on the economy.

[Bob Wallace]

I've now got a copy of the Shaner et al. paper and I've given it a scan.

My first impression it's another "Here's why wind and solar come up short"  paper as opposed to a "How could we get most of our electricity from wind and solar at an affordable price" approach.

Present a lot of data how the Sun sets at night, the wind doesn't always blow, the wind blows more during some times of the year than others, yada, yada, yada.  Model in a modest amount of wind and solar then end up with the conclusion that it would take very large amounts of storage and/or dispatchable generation. 

(They even suggest using ramping nuclear as a fill-in for wind and solar.  That would really be stupid.)

I've done a 'how to' study.  Start with actual hourly demand and wind/solar production on a large geographical area (California) and systematically ramp up the amount of wind and solar generation while watching 1) hours fully supplied, 2) wind/solar penetration, and 3) cost.

The Shaner paper treats overbuilding pretty much as something to be avoided.  I'm not sure if they do not understand how inexpensive it would be to overbuild or if there's something else going on.

I very often see people recoil in horror (that was a bit of drama) over the idea of overbuilding and curtailing wind and solar.  As if we don't already overbuild coal, gas, and nuclear.  US coal and CCNG plants run only about half the time (CFs between 50% and 60%) which is close to a 2x overbuild.  US gas peakers run 10% or less of the time which is a 10x+ overbuild.  France's nuclear plants have a CF of 77% which is higher than it would be if France didn't dump a lot of surplus generation into other European countries. 

IMO we need to treat this issue as a purchasing agent would solve the problem.  Look at the need.  See what amount of the need might be moved around.  Look at the cost of meeting the need with the low carbon options available.  Figure out the combination of low carbon inputs and storage that reliably meets the need for the lowest cost.  Write up a contract.

[Bob Wallace]

Here's the California Duck on June 25, 2017.  Top line is load.  Second line is unserved load after installed wind and solar took their bites.

Bottom line (above zero) is what unserved load would have looked like if CA had 2x as much solar and 8x as much wind installed that day.



17% of the total 2xS 8xW generation would have not been used.  Curtailed.

That curtailment would have meant a rise in electricity cost from $0.0200/kWh to $0.0215/kWh.  And the unused could probably have been sold for other uses.

[crandles]

17% of the total 2xS 8xW generation would have not been used.  Curtailed.

That curtailment would have meant a rise in electricity cost from $0.0200/kWh to $0.0215/kWh.  And the unused could probably have been sold for other uses.

17% unused results in 7.5% cost rise? Why not 17% cost rise?

[Bob Wallace]

17% of the total 2xS 8xW generation would have not been used.  Curtailed.

That curtailment would have meant a rise in electricity cost from $0.0200/kWh to $0.0215/kWh.  And the unused could probably have been sold for other uses.

17% unused results in 7.5% cost rise? Why not 17% cost rise?

Because I apparently divided 0.02 by 0.93 rather than 0.83 which would have been the correct number.  (1.0 - 0.17).

Good catch, thanks.  Cost with 17% curtailment would rise to $0.024/kWh.

[Bob Wallace]

Let me add, this is just one day.  To make a meaningful statement of the value of overbuilding it would be best to have a few years of data. 

And different regions should be considered independently unless there is a high probability of them being highly connected with transmission.  In the US Northeast, for example, overbuilding wind is likely to play a larger role with hydro doing a lot of fill-in. 

In fact, we need robust models that use actual hour by hour (or smaller time blocks) that allow the amount of wind, solar, other renewables, transmission, and storage to be adjusted for the best mixes for current grids.  And to identify where it might make sense to expand grids.

And we need to abandon our fear of overbuilding wind and solar.  They are cheap.  Only using a quarter of "2 cent" wind or solar boosts the price to only "8 cents", leaving them by far our least expensive low carbon energy sources.

[oren]

Here's the California Duck on June 25, 2017.  Top line is load.  Second line is unserved load after installed wind and solar took their bites.

Bottom line (above zero) is what unserved load would have looked like if CA had 2x as much solar and 8x as much wind installed that day.



17% of the total 2xS 8xW generation would have not been used.  Curtailed.

That curtailment would have meant a rise in electricity cost from $0.0200/kWh to $0.0215/kWh.  And the unused could probably have been sold for other uses.

Love the analysis. And I agree, this should be rolled out as soon as possible everywhere.
Advancing from this - what would be the cost of adding ~10000MWh of storage, cycling daily, to get over the 20:00 peak, as opposed to additional overbuilding of wind to get over that same peak? My guess would be such complementary storage would be more economical as well as more reliable than additional overbuilding.

[Bob Wallace]

I've worked storage into my model.  I need to spend some time going through the nuts and bolts to look for bugs.

I see a role for very short term storage, what is needed for 'second to second' demand/supply matching.  And perhaps we wouldn't need much of that.  With some dispatchable EV charging or other controllable loads the grid could simply switch a lot of charging off as needed. 

Follow the "minutes" slope of the demand curve by curtailing wind/solar.  Use EV charging to soak up short term spikes.  I don't see a need for V2G.  Just charge when there's extra and don't charge when there's not.

I suspect most storage is going to be something like pump-up hydro which can serve as short term storage while also storing a lot of energy for a low cost.  Just make the reservoirs larger.